Intelligent Semiconductor Engineering · Chung-Ang University
I am with the Department of Intelligent Semiconductor Engineering at Chung-Ang University, where I have been an undergraduate research intern since July 2026.
My work so far spans computer vision and applied machine learning — generative data augmentation for small-sample industrial defect detection, multimodal sensor learning, and 3D reconstruction pipelines.
Computer VisionGenerative Augmentation3D ReconstructionApplied ML
0 entries on 0 days in the past year
No entries yet. Daily research notes will appear here, grouped by date.
Effects of Generative-AI Augmentation for Small-Sample Industrial Defect Detection
Under extreme data scarcity (~20 images per class), semantic diversity matters more than quantity: traditional augmentation gives −1.76 mAP, Gemini-2.0-based generative augmentation +1.90, and their combination at an 8× optimum +5.0 mAP across detector families.
An Integrated Preprocessing Pipeline for Model Performance Comparison on a Multimodal Gas Sensor Dataset
Designs a standardized four-stage preprocessing pipeline for the MultimodalGasData benchmark and compares eight models fairly, finding a 22.2%p gap between best and worst and showing that the fusion strategy matters more than model depth.
CarNeRF — AI-Powered Used-Car Dealer Platform
An automated pipeline that turns a seller's one-minute video into FastGS-based 3D reconstruction → YOLOv8 defect detection → LightGBM price prediction → LLM natural-language search. PSNR 31.88 with vanilla 3DGS; about 15× faster after porting to FastGS.
FastGS · COLMAP · SAM · YOLOv8 · LightGBM · FastAPI · RAG · React Native
ETF with AI — LambdaRank-Based Trading System
LightGBM LambdaRank trained on 85 features, integrated with an automated TradingView collector and a Next.js dashboard. Live trading through the Korea Investment & Securities API with cron-based daily prediction and monthly retraining; scaling to 1,000 tickers.
LightGBM LambdaRank · FastAPI · Next.js · Playwright · MySQL · Docker · KIS API
JUMP AI 2025 — 3rd Drug Discovery Challenge
Predicts MAP3K5 IC50 activity from SMILES. Designed 140-dimensional features, tuned five base learners with Optuna, and combined them in a two-track ensemble (SLSQP weighting + quantile matching).
RDKit · LightGBM · XGBoost · Optuna · scikit-learn
Personal Color Diagnosis System
An interactive kiosk web service that diagnoses personal color from a face photo and plays tailored video content. Collected ~50K celebrity reference images; white-balancing with OpenCV skin-tone segmentation; OpenAI prompt routing.
PyTorch · OpenCV · OpenAI API · AWS · Docker · React
Feb. 2026 – Jun. 2026
Undergraduate Teaching Assistant — DSML Vibe Coding
DSML, University of Suwon
Mentored undergraduate students.
Sep. 2025 – Dec. 2025
Research Intern — Hadd Science
Sungkyunkwan University
Managed the website, wrote articles, and worked on AutoCAD 3D modeling.
Jul. 2025 – Jun. 2026
Undergraduate Researcher
Prof. Hongryeol Ahn's Lab, University of Suwon
Computer vision research that led to first- and co-authored papers; also built application prototypes with React / Node.js / MongoDB.
Aug. 2024 – Feb. 2025
Intern, Autonomous Driving Team
AIMMO Inc.
Cleaned vision failure data, reviewed Labellerr annotations, collected training data, and analyzed sensor graphs of detection failures.
Jan. 2022 – Jul. 2023
Military Service — Republic of Korea Marine Corps
Honorably discharged.
Jul. 2026 – Present
Chung-Ang University
Undergraduate Research Intern
Mar. 2021 – Feb. 2027 (expected)
University of Suwon
B.S. in Data Science · GPA 4.33 / 4.5
Mar. 2019 – Feb. 2021
Seongil High School